evaluation of gas-particle partitioning in a regional air quality model

Atmos. Chem. Phys., 16, 15327–15345, 2016www.atmos-chem-phys.net/16/15327/2016/doi:10.5194/acp-16-15327-2016© Author(s) 2016. CC Attribution 3.0 License.

Evaluation of gas-particle partitioning in a regionalair quality model for organic pollutantsChristos I. Efstathiou1, Jana Matejovicová1,2, Johannes Bieser3,4, and Gerhard Lammel1,5

1Masaryk University, Research Centre for Toxic Compounds in the Environment, Kamenice 5, 62500 Brno, Czech Republic2Slovak Hydrometeorological Institute, Jeséniova 17, 83315 Bratislava, Slovak Republic3Institute of Coastal Research, Helmholtz-Zentrum Geesthacht, Max-Planck-Str. 1, 21502 Geesthacht, Germany4German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Weßling, Germany5Max Planck Institute for Chemistry, Multiphase Chemistry Department, Hahn-Meitner-Weg 1, 55128 Mainz, Germany

Correspondence to: Christos I. Efstathiou ([email protected])

Received: 22 January 2016 – Published in Atmos. Chem. Phys. Discuss.: 29 February 2016Revised: 23 September 2016 – Accepted: 9 November 2016 – Published: 9 December 2016

Abstract. Persistent organic pollutants (POPs) are of consid-erable concern due to their well-recognized toxicity and theirpotential to bioaccumulate and engage in long-range trans-port. These compounds are semi-volatile and, therefore, cre-ate a partition between vapour and condensed phases in theatmosphere, while both phases can undergo chemical reac-tions. This work describes the extension of the CommunityMultiscale Air Quality (CMAQ) modelling system to POPswith a focus on establishing an adaptable framework that ac-counts for gaseous chemistry, heterogeneous reactions, andgas-particle partitioning (GPP). The effect of GPP is as-sessed by implementing a set of independent parameteriza-tions within the CMAQ aerosol module, including the Junge–Pankow (JP) adsorption model, the Harner–Bidleman (HB)organic matter (OM) absorption model, and the dual Dachs–Eisenreich (DE) black carbon (BC) adsorption and OM ab-sorption model. Use of these descriptors in a modified ver-sion of CMAQ for benzo[a]pyrene (BaP) results in differentfate and transport patterns as demonstrated by regional-scalesimulations performed for a European domain during 2006.The dual DE model predicted 24.1 % higher average domainconcentrations compared to the HB model, which was inturn predicting 119.2 % higher levels compared to the base-line JP model. Evaluation with measurements from the Eu-ropean Monitoring and Evaluation Programme (EMEP) re-veals the capability of the more extensive DE model to bettercapture the ambient levels and seasonal behaviour of BaP.It is found that the heterogeneous reaction of BaP with O3may decrease its atmospheric lifetime by 25.2 % (domain and

annual average) and near-ground concentrations by 18.8 %.Marginally better model performance was found for one ofthe six EMEP stations (Košetice) when heterogeneous BaPreactivity was included. Further analysis shows that, for therest of the EMEP locations, the model continues to under-estimate BaP levels, an observation that can be attributed tolow emission estimates for such remote areas. These findingssuggest that, when modelling the fate and transport of or-ganic pollutants on large spatio-temporal scales, the selectionand parameterization of GPP can be as important as degrada-tion (reactivity).

1 Introduction

Polycyclic aromatic hydrocarbons (PAHs) are a group oflipophilic organic compounds with demonstrated carcino-genicity and potential to bioaccumulate (Diamond andHodge, 2007; Finlayson-Pitts and Pitts, 1999; Pedersen etal., 2004, 2005; WHO, 2003). Typically emitted to the at-mosphere as a mixture of semi- to low-volatile congenerswith variable carcinogenic potency, their fate and transport inthe environment is difficult to assess. Current limitations in-volve inadequate knowledge of photochemistry (Keyte et al.,2013), air–surface exchange (Galarneau et al., 2014; Keyteet al., 2013; Lammel et al., 2009), and combustion sources(Bieser et al., 2012; Lammel et al., 2013). In addition, alu-minium and steel production, along with petrogenic sources,are significant in some environments (Wenborn et al., 1999).

Published by Copernicus Publications on behalf of the European Geosciences Union.

15328 C. I. Efstathiou et al.: Evaluation of gas-particle partitioning in a regional air quality model

This relationship with anthropogenic activity and populationdensity raised awareness of their potential as health stressors,with hotspots identified in urban industrial settings aroundthe world (Šrám et al., 2013; Zhu et al., 2015). On the otherhand, observations of PAHs at remote sites and global cir-culation model studies indicate long-range transport (Keyteet al., 2013; Sehili and Lammel, 2007). Therefore, and dueto resistance to degradation, PAHs are classified as persis-tent organic pollutants (POPs) by the United Nations Eco-nomic Commission for Europe (UNECE) Convention onLong-range Transboundary Air Pollution (CLRTAP) and arelisted in the Protection of the Marine Environment of theNorth-East Atlantic (OSPAR).

Much of the scientific literature is focussed on the 16 USEPA priority PAHs, i.e. acenaphthene (Ace), acenaphthylene(Acy), fluorene (Fln), naphthalene (Nap), anthracene (Ant),fluoranthene (Flt), phenanthrene (Phe), benzo[a]anthracene(BaA), benzo[b]fluoranthene (BbF), benzo[k]fluoranthene(BkF), chrysene (Chr), pyrene (Pyr), benzo[ghi]perylene(BgP), benzo[a]pyrene (BaP), dibenzo[a,h]anthracene(DBahA), and indeno[1,2,3-cd]-pyrene (IPy), albeit withoutjustification (Andersson and Achten, 2015). Used as abiomarker for the carcinogenic risk of PAHs, BaP is acriteria pollutant in many countries, sometimes togetherwith a few other bioaccumulative members. While in NorthAmerica no federal guidelines are currently in effect, theEuropean Union indicative limit value is set at 1 ng m−3

of BaP (EC, 2004), and certain countries have establishedstricter air quality standards (i.e. 0.25 ng m−3 of BaP in theUnited Kingdom) (DETR, 2009).

A number of studies, which aimed at exploring the atmo-spheric fate and transport of POPs on various scales, havebeen presented in the last two decades. These include rela-tively simple mass balance models developed under the La-grangian framework, some of which have been extended tomultimedia applications for studying the accumulation andexchange of POPs between various compartments (Halsall etal., 2001; Lang et al., 2008, 2007; Liu et al., 2007; Preve-douros et al., 2004, 2008; Van Jaarsveld et al., 1997). Eule-rian grid models add yet another level of complexity regard-ing process detail that is accompanied by additional com-putational demands. Global-scale chemical transport mod-els (Friedman and Selin, 2012; Lammel et al., 2009; Sehiliand Lammel, 2007), as well as hemispheric models (Gu-sev et al., 2005; Hansen et al., 2006; Shatalov et al., 2005),have been applied to study the long-range transport of POPs.Moreover, coupled meteorological and air quality modellingframeworks have presented efforts to further investigate thetreatment of physical and chemical atmospheric processesrelated to POPs in higher resolution (Aulinger et al., 2007;Bieser et al., 2012; Cooter and Hutzell, 2002; Galarneau etal., 2014; Inomata et al., 2012; Matthias et al., 2009; Meng etal., 2007; San José et al., 2013; Silibello et al., 2012; Zhang etal., 2009). Table S1 in the Supplement provides a summary ofthe most relevant processes covered in the respective regional

model implementations and subsequent evolution tracked bythe authors. While such studies represent the initial effortsto accurately model POPs on regional scale, important pro-cesses such as gas-particle partitioning, heterogeneous reac-tivity, and a multicompartmental approach for the relevantcompounds are not always studied. The Weather ResearchForecast along with the Community Multiscale Air Qualityfor BaP (WRF-CMAQ-BaP) model presented in this articleis considered to be the first effort to examine the full spec-trum of gas-particle partitioning models relevant to POPs inthe same regional-scale modelling application. The modelattempts to capture POP aerosol chemistry and dynamicswithin the CMAQ framework and can be extended to accountfor mass exchange between adjacent compartments for com-pounds for which such processes are deemed critical (i.e. soilvolatilization).

2 Model development

The model development framework is based on the CMAQversion 4.7.1 modelling system (Byun and Schere, 2006;Byun and Ching, 1999) with the CB05 gas-phase chem-istry mechanism (Yarwood et al., 2005) and the AERO4 ver-sion of the aerosol module (Binkowski, 2003). In addition,CMAQ contains modules representing advection, eddy dif-fusion, and in-cloud and below-cloud scavenging with pre-cipitation. Advection and diffusion satisfy mass conservationand include removal by dry deposition, while in-cloud andprecipitation processes simulate aqueous chemistry and wetdeposition by cloud droplets (Byun, 1999a,b). The aerosolmodule (Binkowski, 2003) determines concentrations of tri-modal size-distributed particulate material with diametersless than 10 µm. According to this model, each mode can bedescribed as an internal mixture of several compounds; themixing state on the single-particle level is not addressed, butis considered equivalent to internal mixing. Aitken and accu-mulation modes include particles with diameters∼ 2.5 µm orless (PM2.5), which are either emitted or produced by gas-to-particle conversion processes. A thermodynamic mech-anism, based on temperature and relative humidity, deter-mines air concentrations of water, sulfate, ammonium, andnitrate aerosols in the fine modes (Nenes et al., 1998; Zhanget al., 2000). Organic and elemental carbon species are alsoincluded in these modes. The third mode represents coarsematerial having diameters between 2.5 and 10 µm and in-cludes dry inorganic material emitted by natural and anthro-pogenic sources. Other processes like coagulation, particlegrowth, aerosol–cloud interactions, and new particle forma-tion are included in the treatment.

CMAQ adaptations are required to fully address the be-haviour of semi-volatile organic compounds (SOCs) in theatmosphere and include additional homogeneous gas-phasereactions, a modular algorithm to describe the mass exchangeof SOCs between the gas and particulate phases, and a de-

Atmos. Chem. Phys., 16, 15327–15345, 2016 www.atmos-chem-phys.net/16/15327/2016/

C. I. Efstathiou et al.: Evaluation of gas-particle partitioning in a regional air quality model 15329

scription of deposition for each phase state of the compoundof interest. For the case of benzo[a]pyrene (BaP), one gas-phase species and three particulate-sorbed species (one foreach aerosol mode) are introduced. It is assumed that only thegaseous molecule undergoes chemical degradation throughreaction with the hydroxyl (OH) radical, leading to an un-specified product that does not undergo further chemical re-action. A summary of the physico-chemical data used in thisstudy is presented in Table 1. Furthermore, a module that ac-counts for heterogeneous chemistry was included for com-pounds for which such reactions represent a significant lossaccording to experimental studies (e.g. BaP reaction withO3).

2.1 Modular implementation of gas-particlepartitioning (GPP) in CMAQ

The distribution of POPs between the gas and the particulatephases is widely recognized as the most important parame-ter in describing their fate and transport in the atmosphere.PAHs like benzo[a]pyrene can be adsorbed to mineral sur-faces or elemental carbon, absorbed by organic aerosol, orabsorbed by aerosol water (Lohmann and Lammel, 2004).Therefore, the aerosol module of CMAQ was enhanced withan array of models that assume different processes to de-termine gas-particle partitioning, i.e. an adsorption model(Junge–Pankow, Junge, 1977; Pankow, 1987), an organicmatter (OM) absorption model (Koa, Harner and Bidleman,1998), and a dual OM absorption and elemental or blackcarbon (EC/BC) adsorption model (Dachs and Eisenreich,2000; Lohmann and Lammel, 2004). In addition, a methodfor calculating the absorption in particulate-phase water wasincluded in the module, employed for all aerosol modes in amanner similar to the GPP processes (Aulinger et al., 2007;Cooter and Hutzell, 2002).

The new GPP algorithm for CMAQ brings together ex-isting implementations for pesticides (Cooter and Hutzell,2002), PAHs (Aulinger et al., 2007; San José et al., 2013),polychlorinated biphenyls (PCBs) (Meng et al., 2007),dibenzo-p-dioxins, and dibenzofurans (PCDD/Fs) (Zhang etal., 2009) in a unified module that allows for flexible GPPmodel combinations, making it applicable to a wide range ofSOCs. The following assumptions were taken into consider-ation: (1) instantaneous relaxation to phase equilibrium (nomass transport kinetic limitations), (2) the compound doesnot irreversibly transform in the particulate phase, and (3)sorption processes do not interact. By extending the gas-phase and modal aerosol treatments in CMAQ for SOCs,these assumptions allow the use of simple ratios to determinethe concentration of each form, following

ai = φi (ai + g), (1)

where ai is the particulate concentration in each mode (i =1=Aitken, 2= accumulation, 3= coarse), g is the gaseousconcentration, and φi is the particulate fraction of the com-

pound in mode, with a general formulation (Aulinger et al.,2007; Cooter and Hutzell, 2002).

φi = φadi +φ

abi +φ

aqi (2)

This way, all partitioning processes, adsorption (φadi ), ab-

sorption in OM (φabi ), and absorption (solution) in aerosol

water (φaqi ), are simultaneously treated. Following the rule

of mass conservation for the compound in both phases, thetotal concentration ctot can be introduced:

g = ctot− a1− a2− a3, (3)

leading to a system of three linear equations. 1 φ1 φ1φ2 1 φ2φ3 φ3 1

q a1a2a3

= ctot×

φ1φ2φ3

, (4)

which yield the particulate concentrations ai in each mode.Subsequently, using the previous equation, the gaseous con-centration g can be calculated.

2.1.1 Junge–Pankow adsorption model

The first GPP scheme is based on the model of exchangeableSOC adsorption to aerosols presented by Junge (1977) andlater critically reviewed by Pankow (1987).

φadi = cJSi/

(pL◦+ cJSi

)(5)

The equation above relates the mass fraction of chemical ad-sorbed particles in each mode i (φad

i ) to the subcooled liquidvapour pressure of the compound (pL

◦, Pa) and the aerosolsurface density (Si , m2 of particle surface per m3 of air). Theparameter cJ (unit, Pa m) depends on the chemical nature ofboth the adsorbent and the adsorbate and cannot be expectedto be a constant, as confirmed by field experiments (Lammelet al., 2010). Nevertheless, a value of cJ = 0.172 Pa m hasbeen suggested (Pankow, 1987) and is often used in mod-elling applications. The surface area parameter, Si , which isa characteristic of the aerosol required for the Junge–Pankow(JP) model, is calculated at each time step for each modewithin the aerosol module of CMAQ (Binkowski, 2003).

The empiric JP model, despite assuming adsorption to bethe only relevant partitioning process, may eventually not befree from absorptive contributions (to cJ ), but these shouldbe very low. The fitting in the empiric parameterization wasdone for all possible vapours, many of these not solublein octanol. No cJ for BaP derived from ambient measure-ments has ever been published. However, we recently foundin background aerosol of central Europe (Košetice 2012–2013, n= 162 aerosol samples, cJ derived from experimen-tal S/V data; Shahpoury et al., 2016) that cJ for BaP is verylow, at 0.01–5.46 Pa cm (median 0.44 Pa cm), supporting theperception that a possible implicit absorptive contribution tocJ must be very low.

www.atmos-chem-phys.net/16/15327/2016/ Atmos. Chem. Phys., 16, 15327–15345, 2016


Table 1. Physicochemical parameters and reaction rate coefficients used in the WRF-CMAQ-BaP model.

Inverse Henry’s law constant1 log H ′ −4.7Subcooled liquid vapour pressure1,2 log pL −5.2 (Pa)Octanol–water partition coefficient1 log Kow 5.9Octanol–air partition coefficient1,3 log Koa(dry) 11.11 (b = 12.04, m= 5382)3

Water-saturated octanol–air partition coefficient1 log Koa(wet) 10.6Diesel soot–water partition coefficient4 log Ksoot-water 8.4Diesel soot–air partition coefficient4,8 log Ksoot-air 13.04, 11.598

Gas-phase OH reaction rate constant kOH 50× 10−12 (cm3 mol−1 s−1)

Ozone gas–surface equilibrium constant6 KO3 0.028× 10−13 (cm3)

Maximum rate coefficient6 kmax 0.060 (s−1)

1 Beyer et al. (2000). 2 de Maagd et al. (1998). Offenberg and Baker (1999). 3 Odabasi et al. (2006) log Koa =mT−1 (K)+ b.4 Bärring et al. (2002), log Ksoot-water = 1.39 log Kow+ 0.1. 5 log Ksoot-air = log Ksoot-water − log H ′. 6 Estimated values(EPIWIN;USEPA). 7 Kwamena et al. (2004). 8 van Noort (2003).

2.1.2 Koa absorption model

The second GPP scheme follows the work of Pankow (1994,2007), with refinements from Harner and Bidleman (1998)that described how the particulate fraction of a compoundcould be derived from the aerosol–air partitioning coefficientKp. Following this approach for each mode i, the predictedparticulate fraction φab

i is linked to the total mass concentra-tion of suspended particles cTSPi with the following equation:

φabi =K

OMp,i cTSPi/

(KOMp,i TSPi + 1

). (6)

Pankow (1987, 1998) demonstrated that the gas-particle par-titioning coefficient, Kp, and the use of pL

◦ as its descriptorare valid both for adsorption and absorption of SOCs. In thecase of absorptive partitioning, Pankow (1998) derived

KOMp,i

(m3µg−1

)=

fOMi760RT

106MOMγOMpL◦, (7)

where fOMiis the mass fraction of OM in the particle

that can absorb gaseous SOCs, R is the ideal gas constant(= 8.314 Pa m3 mol−1 K−1), T (K) is absolute temperature,MOM (g mol−1) is the mean molecular weight of the OMphase, γOM is the activity coefficient of the selected SOC inthe OM on a mole fraction basis, and pL

◦ (Pa) is the vapourpressure of the pure SOC (subcooled liquid in the case ofsolids). Partitioning coefficients normalized to the OM con-tent were introduced to further support the absorption hy-pothesis (Finizio et al., 1997). Pankow (2007) and Harnerand Bidleman (1998) replaced pL

◦ by the octanol–air parti-tioning coefficient, Koa.

KOMp,i = 10−12 fOMi

MoctγoctKoa

MOMγOMρoct, (8)

whereMoct,MOM (g mol−1) are the molecular weights of oc-tanol (= 130 g mol−1) and the OM phase, γoct, γOM are theactivity coefficients of the SOC in octanol and OM, respec-tively, and ρoct is the density of octanol (0.82 kg L−1). As-suming that octanol imitates organic matter in PM, Harner

and Bidleman (1998) suggested that the ratio of γoct/γOMand Moct/MOM can be assumed to be 1. However, it waslater suggested that MOM could be much higher, particularlyin secondary organic aerosols containing polymeric struc-tures (Kalberer et al., 2004); a mean value of 500 g mol−1

was later suggested by Götz et al. (2007), which results inMoct/MOM of 0.26. Under the assumption that γoct/γOM andMWoct/MWOM = 1, the equation above can be simplified tothe following:

logKOMp,i = logKoa+ logfOM− 11.91. (9)

2.1.3 Dual OM absorption and black carbonadsorption model

Finally, the third GPP scheme extends the previous OM ab-sorption model to include adsorption onto particulate soot,according to evidence showing the measured Kp to be ex-ceeding the predicted Kp based solely on absorption inOM (Dachs and Eisenreich, 2000; Fernandez et al., 2002;Lohmann and Lammel, 2004; Ngabe and Poissant, 2003).Considering the contribution of those two additive processes,a modified partition coefficient can be employed.

Kdualp,i = 10−12

(fOMi

MoctγoctKoa

MOMγOMρoct+fBCi aatm-BCKSA

asootρBC

), (10)

where fBCi is the fraction of black carbon in the particulatematter, aatm−BC and asoot are the available surfaces of at-mospheric black carbon (BC) and diesel soot, respectively,KSA is the partitioning coefficient between diesel soot andair. A density value of 1 kg L−1 has often been assumed forBC; however values in the range of 1.7–1.9 kg L−1 have beenmeasured and recommended by Bond and Bergstrom (2006).Thus, the equation can be simplified.

Kdualp,i = 10−12 (0.32 × fOMi

Koa+ 0.55 × fBCiKSA). (11)

This method is subject to uncertainties (Goss, 2004), butis accepted and suitable (Dachs et al., 2004 and references



therein). Soot–air partition coefficients (KSA) were calcu-lated as the ratio of soot–water adsorption constants KSWand the inverse Henry’s law constant (H ′), with KSW val-ues adopted from Bärring et al. (2002). As a step to addressuncertainties contributed by different methods in calculatingKSA, a thermodynamic estimation model suggested by vanNoort (2003) was also tested:

logKSA =−0.85logpL◦+ 8.94− log(998/asoot). (12)

The soot-specific surface area was set to 18.21 m2 g−1, de-rived as the geometric mean of surface areas for traffic, wood,coal, and diesel soot (i.e. 59.4, 3.6, 8.2, and 62.7 m2 g−1,respectively) (Jonker and Koelmans, 2002). Additionally, inorder to estimate uncertainties arising from Koa and its tem-perature dependence, two different parameterizations weretested, the first based on the work of Beyer et al. (2000) andthe second following the temperature-dependent parameteri-zation suggested by Odabasi et al. (2006).

2.2 Dissolution into aerosol water

An assumption of the CMAQ aerosol module is that organicsinfluence neither the water content nor the ionic strength ofthe system. While this assumption may not be valid for manyatmospheric aerosols, sufficient basic data are not availableto treat the system in a more complete way. In a similar man-ner, the dissolution of PAHs in the aqueous electrolyte wascalculated using the inverse Henry’s law constant (Aulingeret al., 2007; Cooter and Hutzell, 2002; Lohmann and Lam-mel, 2004). Lohmann and Lammel (2004) derived the parti-tioning coefficient Kaq

p,i , although they conclude that as longas a solid–air interface exists, the contribution of dissolvedPAHs to the gas-particle partitioning will always be negli-gible. To account for this, a “wet aerosol” switch was in-troduced and associated with the relevant partitioning coeffi-cient.

logKaqp,i = logKwa+ logfwa,i − 12.0, (13)

where fwa,i is the mass fraction of aerosol water in eachmode i and Kwa is the water–air partitioning coefficient,which is equal to H ′. The aerosol water content is computedin CMAQ based on a variation of the Zdanovskii, Stokes,and Robinson (ZSR) method (Byun and Ching, 1999; Kimet al., 1993; Stokes and Robinson, 1966). As suggested byAulinger et al. (2007) for CMAQ, only if the ratio of am-monium sulfate to aerosol water drops below the solubilityof ammonium sulfate, the aerosol is treated as wet and ab-sorption into aerosol water takes effect, while adsorption toinorganic material is switched off.

2.3 Gas-phase and heterogeneous reactions

Reactions with the hydroxyl radical (OH), ozone (O3), andthe nitrate radical (NO3)may determine the atmospheric fateand long-range transport potential of SOCs. Gas-phase re-actions with OH are believed to be the dominant loss path-way for semi-volatile PAHs (3–4 rings) like anthracene andpyrene (Atkinson and Arey, 1994; Finlayson-Pitts and Pitts,1999; Keyte et al., 2013). On the other hand, in the case ofnon-volatile PAHs (5 and more rings, like BaP), loss due toheterogeneous reaction with ozone is believed to dominate(Finlayson-Pitts and Pitts, 1999). Experimental studies of thekinetics of PAHs sorbed to particles indicate strong influ-ences of mixing state (single-particle anisotropy), morphol-ogy, and phase state of the particles (Kwamena et al., 2004;Perraudin et al., 2007; Pöschl et al., 2001; Shiraiwa et al.,2009; Zhou et al., 2012). Yet, detailed heterogeneous chem-istry models would require corresponding additional param-eters and a more sophisticated surface layer model (Kaiseret al., 2011; Shiraiwa et al., 2010; Springmann et al., 2009).However, in the studies by Kahan et al. (2006) and Kwamenaet al. (2004, 2007), the reactivity could be well described bya Langmuir–Hinshelwood mechanism (i.e. one species [BaP]is adsorbed to the surface while the second species, ozone,is in phase equilibrium). This approach is followed in theCMAQ aerosol module by introducing the degradation ratecoefficient k in the following form:

k =kmaxKO3cO3

1+KO3cO3

, (14)

where kmax is the maximum rate coefficient (0.060± 0.018s−1), KO3 is the ozone gas to surface equilibrium constant(0.028± 0.014× 10−13 cm3), and cO3 is the concentration ofozone (mol cm−3). As this reaction neglects the substancefraction not accessible for gaseous O3 molecules (so-calledburial effect; Zhou et al., 2012), an upper estimate for BaPreactivity is simulated.

2.4 Dry and wet deposition

SOCs and their degradation products are removed from theatmosphere through dry and wet deposition. Dry depositionis modelled in CMAQ as a one-way flux out of the lowestlayer of the atmosphere. Particulate forms have dry depo-sition velocities determined by Brownian diffusion, turbu-lence, and sedimentation. Mass and diameter of the modeare used to estimate the sedimentation velocity (Binkowski,2003; Byun and Ching, 1999). The deposition velocity ofgaseous molecules is based on an electrical resistance ana-logue that includes aerodynamic (Ra), quasi-laminar bound-ary layer (Rb), and surface canopy (Rc) resistance (Pleimet al., 1999). Ra is computed using similarity theory withheat flux. Rb depends on land-use-specific friction velocityand molecular characteristics of gases. Rc consists of severalcomponents, including bulk stomatal, dry cuticle, wet cuti-



cle, ground, and in-canopy aerodynamic resistance (Pleimet al., 1999). Cuticle and ground resistances are based onozone and sulfur dioxide observations and neglect absorptioninto organic plant or soil material. Values for time-dependentRc parameters are determined via land use and mesoscalemeteorological models, which provide fractional vegetationcover, leaf area index, fractional leaf area wetness, and leafstress associated with radiation, root zone soil moisture, tem-perature, and humidity (Pleim and Xiu, 1995).

Wet deposition of gaseous molecules occurs through in-cloud and below-cloud scavenging. Both processes are as-sumed to have the same scavenging efficiency and depen-dency on Henry’s law, water content of the cloud, and precip-itation rate (Chang et al., 1986). Wet scavenging parameteri-zation of the particulate-phase molecules follows the methodof Roselle and Binkowski (Byun and Ching, 1999) and de-pends on the aerosol mode (size) and time required to removeall liquid water from a cloud volume. No ice-phase scaveng-ing is included in the fourth version of the aerosol module inCMAQ.

3 Model implementation and study design

The model domain included the whole of Europe and neigh-bouring countries as shown in Fig. 1. The domain is withinthe area covered by the UNECE-CLRTAP (and its POP pro-tocol, covering PAHs) and was specifically chosen to ex-ploit recently prepared high-resolution emission data (Bieseret al., 2011) and make use of the only monitoring networkin the region (EMEP; see below). Computational simula-tions were performed on a 36× 36 km Lambert ConformalConic grid using offline hourly meteorological fields, gen-erated by the Weather Research Forecasting (WRF) modelversion 3.4.1 using GFS meteorological reanalysis data (spa-tial resolution 0.5◦× 0.5◦) as input. The parameterizationof WRF included the following schemes: Milbrandt–Yaudouble-moment 7-class microphysics (Milbrandt and Yau,2005a, b), rapid radiative transfer model (RRTMG) long-wave and shortwave scheme (Iacono et al., 2008), asymmet-ric convective model of the planetary boundary layer (PBL)(Pleim, 2007), Pleim–Xiu surface layer model (Pleim andXiu, 2003; Xiu and Pleim, 2001), and the improved versionof Grell–Devenyi (Grell, 1993; Grell and Devenyi, 2002) en-semble scheme for cumulus parameterization. On the verti-cal dimension, the domain was resolved in 36 layers, fol-lowing the eta coordinate system. Subsequently, CMAQ-ready meteorological input fields for 2006 were prepared us-ing the Meteorology-Chemistry Interface Processor (MCIP)(Otte and Pleim, 2010). Computational simulations were per-formed in a distributed computing infrastructure operatedand managed by the Czech National Grid Organization Meta-Centrum NGI.

Figure 1. Study domain and 2006 annual grid cell mean BaP emis-sion flux (g s−1) at the surface.

3.1 Emissions

Capturing the spatio-temporal pattern of emissions is cru-cial to determine the fate and transport of PAHs. Severalattempts have been carried out during the past decades,with varying spatial and temporal resolutions in historical,current, and future projection terms for Europe (Bieser etal., 2012; Pacyna et al., 2003). Currently, four databasesources with gridded BaP emissions exist for Europe andRussia: the TNO database (Denier van der Gon et al., 2006),the EMEP database (Mantseva et al., 2004), a dataset forPOPCYCLING-Baltic project (Pacyna et al., 2003), and aglobal emission inventory for the year 2008. The SMOKEfor Europe (SMOKE-EU) model (Bieser et al., 2011) out-put used in this study follows a methodology that is basedon the TNO database and has been evaluated in a number ofsubsequent CMAQ applications (Aulinger et al., 2011; Bew-ersdorff et al., 2009; Matthias et al., 2008). In order to spa-tially disaggregate BaP emissions in a more realistic way,SMOKE-EU improves on using a linear dependency on thepopulation density as surrogate factor by introducing differ-ent relevant relationships based on source classification andmeteorological parameters. For example, the relationship towood availability and the level of urban agglomeration havebeen introduced in an effort to include the effect of burningbiomass. The resulting spatial distribution and annual aver-age emission flux of BaP for our domain during the year 2006is shown in Fig. 1. In addition, BaP emissions are shapedtemporally using diurnal and weekly functions. As residen-tial combustion is associated with the majority of BaP emis-sions, a strong dependence on the season and spatial vari-ability is expected. To account for this dependence, heating



Figure 2. Annual average surface-level BaP concentrations [log cBaP (ng m−3)] during 2006 plotted in logarithmic scale for (a) the Junge–Pankow (JP) scenario 1 and (b) for the fully expanded GPP scenario 5, which includes a heterogeneous reaction with ozone.

degree days have been introduced as a proxy (Aulinger et al.,2011). A major strength of the SMOKE-EU inventory is theprovision of CMAQ-ready emissions for all species involvedin the desired chemical mechanism (e.g. CB5) and aerosolmodule (e.g. aero4). Following the trimodal representationof aerosol in CMAQ, 0.1 % of the emissions of BaP was al-located to the Aitken mode, while the rest was emitted in theaccumulation mode, as this is the default treatment with allprimary organic aerosol species in the model.

3.2 Evaluation data and sensitivity simulations

The spatio-temporal coverage of BaP in air and depositionin Europe and Russia is limited, in many regions sporadic ornot available at all. For 2006, concentration in air is availablefrom eight stations (n= 173), while deposition informationis available from six stations of the EMEP monitoring net-work (Aas and Breivik, 2012). An overview of the data, in-cluding site codes, geographical information, and the sam-pling strategies, is provided in Table S2. It is evident thatduring 2006, only 4 of the total 10 sites provide informationon concentration and deposition simultaneously. Addition-ally, the sampling strategy varies widely among the stations,leading to a non-homogeneous pool of data. For example, atthe Košetice (CZ0003R) and Niembro (ES0008R) stations,BaP is measured for 24 h once a week, while in Aspvreten(SE0012R) and in Pallas (FI0036R) samples cover one weekand are taken once every month. On the other hand, the Lat-vian sites (LV0010R, LV0016R) are sampled on a monthlybasis, while for the High Muffles location (GB0014R) a 3-month sample is obtained four times in 2006. Finally, the

sites located in Germany (DE0001R, DE0009R) provided in-formation only in terms of wet deposition and on a monthlybasis during 2006.

Based on the set of GPP mechanisms described before, theeffect of each parameterization was evaluated with the EMEPmeasurements by incrementally testing the models involved,while keeping the same meteorological and emission inputs.Table 2 summarizes the five annual scenarios that were devel-oped for 2006: (1) Junge–Pankow, (2) scenario 1 with waterdissolution, (3) scenario 2 with the Harner-Bidleman (HB)Koa absorption model, (4) scenario 2 with dual OM absorp-tion and black carbon adsorption model Dachs-Eisenreich(DE) scheme, and (5) scenario 4 supplemented by a hetero-geneous reactivity module. A spin-up time of 7 days wasallowed for each simulation in order to maintain a low in-fluence of the initial conditions. Comparisons with measure-ments of BaP in ambient air were produced using the sta-tistical computing framework of the R language, and themodel performance was evaluated, building on a wide rangeof statistics calculated using the Models-3 and openair pack-ages (Carslaw and Ropkins, 2012).

4 Results and discussion

4.1 Spatio-temporal distributions of BaP over Europe

The pattern of normalized annual mean BaP concentrationsover the domain using the baseline scenario 1 (JP) is pre-sented in Fig. 2a. Elevated levels were found in regions withstrong emission sources in central Europe (Po Valley, Rhine-Ruhr, Poland, and Ukraine areas), as well as in the vicinity



Table 2. Summary of the simulated scenarios designed to test gas-particle partitioning and heterogeneous reactivity of BaP.

Scenario

Description and references 1 2 3 45 5

Gas-phase reaction with OH Yes Yes Yes Yes YesAdsorption model (JP)1 JP JP JP JP JPAerosol water dissolution (W) – W W W WAbsorption to OM (HB)2 – – HB – –Absorption to EC/BC (DE)3 – – – DE DE

Heterogeneous reaction with O3 (KW)4 – – – – KW

1 Junge (1977) and Pankow (1987). 2 Harner and Bidleman (1998). 3 Dachs and Eisenreich (2000).4 Kwamena et al. (2004). 5 Additional simulations with temperature-dependent octanol–air partitioncoefficient calculated based on Odabasi et al. (2006) and a thermodynamic estimation modelsuggested by van Noort (2003).

of large cities (Moscow, Paris, Vienna, Madrid, and Istan-bul, among others). In addition, annual mean BaP concentra-tions according to the fully expanded scenario 5 (KW) aredepicted in Fig. 2b. Comparing the output from the afore-mentioned simulations revealed higher BaP levels with sce-nario 5, which can be associated with the geographic dis-tribution of emissions. This is the first indication that thechosen partitioning approach has a significant impact on themodel results for total BaP concentrations. However, it is alsoevident that in southern Europe and the Mediterranean re-gion, the BaP losses associated with the addition of an up-per estimate of heterogeneous reactivity with O3 (scenario 5,Fig. 2b) may have a limiting effect on BaP long-range trans-port (LRT) potential. This can be supported by the promi-nent BaP gradients calculated over the Mediterranean Sea,acting as a sink that is driven by low BaP emissions and el-evated O3 levels. The overall model patterns show a goodagreement with recent regression methods mapping BaP lev-els that provide with a partial coverage of our EU domain.With respect to previous studies involving simulations of BaPover Europe, we find lower BaP concentrations, but withinthe same order of magnitude as those estimated by initialCMAQ applications (Aulinger et al., 2007; Bewersdorff etal., 2009; Matthias et al., 2009). This supports the perceptionthat BaP emissions over Europe have decreased substantiallyduring the recent decades, a feature improved in subsequentevolution of the emissions model which led to better agree-ment with the more recent CMAQ applications (Bieser et al.,2012). A closer look over Italy and the Iberian Peninsula re-veals a good agreement with previous studies depicting BaPdistributions for these regions (San José et al., 2013; Silibelloet al., 2012). Finally, discrepancies over eastern Europe indi-cate that a more elaborate emissions inventory (i.e. domes-tic heating adjustments, fire activities) would be beneficial toaddress the elevated BaP levels predicted by recent hybridmodelling approaches (Guerreiro et al., 2016a).

4.1.1 Effect of GPP parameterizations andheterogeneous reactivity with O3

Figure 3 and Table 3 illustrate the differences resulting fromthe incrementally tested scenarios (Table 2), expressed as an-nual mean BaP concentrations over the entire domain of in-terest and the individual 28 EU member states. As expected,the effect of dissolution into aerosol water (Fig. 3a) is limitedthroughout most of the domain, with coastal urban areas be-ing the most influenced (i.e. Istanbul, North Sea). Accordingto scenario 2, BaP dissolution in aerosol liquid water led to areduction of less than 0.36 % of ground-level BaP in Europecompared to the baseline JP scheme (Table 3). Introducingthe HB absorption scheme (scenario 3) resulted in∼ 119.2 %higher average BaP concentrations (Fig. 3b) with extremesobserved in areas associated with emission sources in cen-tral Europe (i.e. Po Valley) and large cities (i.e. Moscow,Paris, Vienna, and Istanbul). Expressed in terms of domain-wide atmospheric lifetimes (defined as τ = btot/Fem, wherebtot denotes the total atmospheric burden and Fem the totalemissions), scenario 3 leads to a 210 % increase for BaP overscenario 2. Similarly, shifting to the DE scheme (scenario 4)appears to have an additive effect on near-ground BaP con-centrations with total lifetimes reaching a domain-average in-crease of 13 %. Compared to scenario 3, scenario 4 affects alarger area of the domain (Fig. 3c), even though the magni-tude of the effect is less pronounced due to its associationwith the availability of black carbon. The domain mean ofmodelled elemental carbon is 0.0212 µg m−3 and the spatialdistribution matches its emission pattern (Fig. S3). These lev-els are significantly lower than in the field studies in the ur-ban air of south-western Europe (EEA, 2013; Milford et al.,2016) and may subsequently lead to a hindered performanceof the DE scheme.

Finally, the introduction of heterogeneous chemistry (sce-nario 5) had, as expected, a substantially negative effect onthe BaP concentrations within our domain, culminating inmajor conurbations (i.e. Po Valley, Moscow, and Istanbul;



Figure 3. Differences in annual average surface-level BaP concentrations (ng m−3) during 2006 plotted for (a) water dissolution effect:scenario 1 to scenario 2, (b) Koa effect: scenario 3 to scenario 2, (c) dual model effect: scenario 4 to scenario 3, and (d) O3 degradationeffect: scenario 4 to scenario 5.

see Fig. 3d). Heterogeneous degradation due to the reactionwith O3 accounts for an approximate 18.8 % reduction of themean ground-level BaP and a 25 % reduction of its atmo-spheric lifetime in the European domain (Table 3). As men-tioned in Sect. 4.1, the BaP concentration gradients reveal astrong spatial link with emission sources in the south, as wellas high O3 concentrations typically found along the coastsof the Mediterranean Sea (Fig. S4). CMAQ calculates ozonepatterns that are in accordance to the findings of recent analy-ses, exhibiting a similar performance to measurements at theEuropean level (de Smet et al., 2009).

4.1.2 Seasonality and relevant parameters affectingBaP fate and transport

Seasonally disaggregated distributions of BaP, as simulatedin scenario 4, are presented in Fig. 4. Strong emissions dur-ing wintertime and autumn have an apparent effect on the av-erage ground-level concentrations. In addition, a normalizedeffect of long-range transport for BaP can be observed in thevicinity of Istanbul and other coastal metropolitan areas, overparts of the large water bodies of the Mediterranean and theBlack Sea. This scenario, however, does not include the treat-ment of the heterogeneous reaction with ozone and, there-fore, provides an upper-bound estimate of the LRT potential



Table 3. Average ground-level BaP concentrations, cBaP (pg m−3) over the entire domain and in individual 28 EU member states for allsimulated scenarios and concentration changes 1cBaP (%) between subsequent scenarios.

1 2 3 4 5

cBaP Conc. cBaP 1cBaP cBaP 1cBaP cBaP 1cBaP cBaP 1cBaPpg m−3 pg m−3 % pg m−3 % pg m−3 % pg m−3 %

Domain average 10.4 10.4 −0.4 22.8 119.2 28.3 24.1 23.0 −18.8Austria 58.4 58.4 0.0 89.0 52.4 103.4 16.1 86.5 −16.3Belgium 36.2 36.1 −0.1 65.8 82.1 75.8 15.2 62.5 −17.6Bulgaria 14.7 14.7 −0.1 28.9 97.1 37.7 30.4 26.7 −29.3Croatia 15.2 15.2 −0.1 35.1 131.0 45.5 29.9 32.2 −29.4Cyprus 11.2 11.2 0.0 13.2 17.8 24.7 87.2 10.1 −59.1Czech Republic 38.6 38.6 0.0 70.1 81.5 83.5 19.1 67.9 −18.7Denmark 7.3 7.2 −1.3 22.4 213.0 26.0 16.0 24.5 −5.7Estonia 9.5 9.5 −0.9 28.9 205.3 34.6 19.7 32.0 −7.4Finland 6.0 5.8 −2.9 16.1 176.0 21.9 36.2 17.7 −19.5France 27.1 27.1 0.0 48.0 76.9 56.6 18.0 44.7 −21.0Germany 30.3 30.3 −0.1 57.6 90.2 69.2 20.2 55.6 −19.6Greece 7.4 7.4 −0.1 14.7 97.7 22.3 52.2 12.4 −44.6Hungary 29.3 29.3 −0.1 61.3 109.5 75.1 22.5 59.7 −20.5Ireland 3.8 3.8 −0.1 6.8 80.0 8.7 27.5 6.0 −31.7Italy 36.9 36.9 0.0 62.8 70.1 76.3 21.6 57.9 −24.1Lithuania 10.0 10.0 −0.5 32.7 228.4 38.4 17.5 36.3 −5.5Luxembourg 37.1 37.1 0.0 62.0 67.2 71.7 15.8 57.9 −19.3Latvia 6.0 5.9 −1.3 25.7 335.0 30.2 17.7 29.8 −1.4Malta 0.5 0.4 −0.5 0.5 20.6 0.7 37.1 0.5 −36.6Netherlands 24.0 23.9 −0.3 52.5 119.5 63.1 20.2 51.5 −18.4Poland 25.6 25.6 −0.2 56.7 121.7 67.4 18.8 57.4 −14.8Portugal 15.0 15.0 0.0 20.9 39.1 28.4 36.3 18.1 −36.5Romania 12.7 12.7 0.0 27.4 115.8 36.0 31.4 25.7 −28.6Spain 8.4 8.4 0.0 11.9 41.8 16.8 41.4 9.3 −44.8Sweden 3.7 3.6 −2.4 11.3 211.7 14.3 26.7 13.2 −7.7Slovakia 27.0 27.0 −0.1 53.5 98.1 65.2 21.9 52.5 −19.5Slovenia 28.5 28.5 0.0 60.3 111.3 71.3 18.3 57.0 −20.1United Kingdom 3.3 3.3 −0.4 8.9 170.6 11.4 28.4 8.2 −27.6

of BaP. On the other hand, Fig. 5 demonstrates the seasonaleffect of the heterogeneous reaction of BaP with ground-levelozone on BaP concentrations. Distributions of BaP concen-tration differences reveal a stronger reduction during winterthat is influenced by central and southern European emis-sion sources and midlatitude ozone availability. Interestingly,the differences between winter and summertime appear to becomparable in terms of magnitude, despite the substantiallyweaker sources during summer that shift the process to thesouth.

An important factor for the performance of the BaP modelis the mass concentration and composition of the aerosolsin the three different modes and their relationship withthe temperature of the ambient air. Figure 6 illustrates theaerosol distributions as simulated at model cells represent-ing three different settings (urban: the city of Vienna, sub-urban: CZ0003R, remote: FI0036R) during a typical winterand summer day. As we move further away from the vicinityof the sources, we notice not only a decrease of the over-

all BaP levels but also a change in the distribution influ-enced by each individual simulated scenario. While the ba-sic Junge–Pankow implementation results in higher coarse-mode BaP concentrations, the HB and DE schemes accentu-ate the strength of the accumulation mode where most of theBaP aerosol mass is emitted. The disproportionately highercoarse-BaP fractions calculated at the remote grid cells dur-ing the summertime can be explained by conditions favour-ing long-range transport.

Regarding the parameterizations of the HB and DEschemes, the calculation of the parameter Koa has a di-rect effect on the relevant model results (scenarios 3and 4). Use of the regression parameters for the semi-logarithmic temperature-dependent form proposed by Od-abasi et al., (2006) results in a mixed effect over its prede-cessor scheme, which is seasonally disaggregated and pre-sented in Fig. S5. These figures suggest that the partitioningbased on eachKoa approach is sufficiently similar for most ofthe domain during the winter. However, differences between



Figure 4. Seasonally disaggregated average surface-level BaP concentration [log cBaP (ng m−3)] maps predicted by scenario 4 during 2006.

the two expressions with respect to simulating summertimeBaP concentrations over central Europe were noted. In ad-dition, simulations that address the uncertainties contributedby the method for estimating KSA were performed for thewinter period of 2006 using the DE scheme. Figure S6 re-veals similar results, with BaP levels slightly higher in cellswith strong emissions sources. However, with the exceptionof Moscow, the average difference BaP concentration in ur-ban cells did not exceed 0.02 ng m3 whenKSA was calculatedbased on the thermodynamic estimation model suggested byvan Noort (2003). Similarly to Koa, results suggest that thepartitioning based on each KSA approach is sufficiently sim-ilar.

4.2 Model evaluation with EMEP measurements

The EMEP network database and model output from all fivescenarios were imported in the R language framework to cre-ate measurement–model pairs for further analysis. Figure 7illustrates the resulting time series plots for selected EMEPsites against CMAQ output extracted from scenario 4. Thisfigure also demonstrates differences in terms of samplingprotocols among individual sites, with the extreme of theHigh Muffles station (4 samples year−1 – Table S2). Aggre-gate temporal profiles of the modelled and measured BaP

concentrations across all sites depicted in Fig. 8 reveal astrong seasonal pattern that is captured well by the model.However, BaP concentrations appear to be underestimatedduring the colder months. This effect can be attributed tothe strength of emissions, which is probably related to inad-equate coverage of residential heating in current inventories(wood/biofuel burning). Figure S7 illustrates this by com-paring annual average BaP emission fluxes between modelcells enclosing EMEP monitoring site locations and selectedmetropolitan areas of Europe during 2006. While the defi-ciency of PAH inventories is known, it may limit the use-fulness of background measurements (such as those in theEMEP network) in assessing the overall model performanceunder different GPP scenarios.

It is evident from the metrics obtained across all sites (Ta-ble 4 and Taylor diagram summary Fig. S8) that the increasedcomplexity in GPP formulation results in better agreementwith the EMEP measurements. For reasons mentioned inSects. 2.2–2.3 and 4.1.1, deviations from this observationare the simulated scenarios that introduce the dissolution toaerosol water (scenario 2) and degradation due to the hetero-geneous reaction with ozone (scenario 5). Overall, the dualmodel (DE) employed in scenario 4 showed the best agree-ment with the measurements. The DE parameterization alsoperformed best when compared with the JP and HB schemes



Table 4. Comparison between modelled and measured BaP concentrations for the entire 2006 EMEP dataset (n= 173) along with perfor-mance metrics and ranking across all sites.

Rank Scenario FAC2 MB MGE NMB NMGE RMSE r COE IOA

1 4 0.29480 −0.05622 0.07145 −0.65264 0.82950 0.14729 0.35193 0.22443 0.612222 3 0.25434 −0.06279 0.07280 −0.72889 0.84516 0.15081 0.33423 0.20979 0.604903 5 0.21387 −0.07021 0.07345 −0.81511 0.85272 0.15179 0.47598 0.20273 0.601364 1 0.18497 −0.07640 0.07887 −0.88699 0.91559 0.16069 0.31088 0.14395 0.571975 2 0.17919 −0.07649 0.07894 −0.88798 0.91642 0.16076 0.30793 0.14317 0.57158

FAC: fraction of predictions within a factor of two, MB: mean bias, MGE: mean gross error, NMB: normalized mean bias, NMGE: normalized mean gross error,RMSE: root mean squared error, r: correlation coefficient, COE: coefficient of efficiency, IOA: index of agreement.

Figure 5. Average difference of surface-level BaP concentrations(ng m−3) between scenario 4 and the fully expanded GPP sce-nario 5, which includes the heterogeneous reaction with ozone dur-ing the winter and summer months of 2006 (scenario 4–scenario 5).

on global-scale applications (Lammel et al., 2009). How-ever, when looking at site-specific performance metrics (Ta-ble S9), the Košetice location (CZ0003R) reveals a betteragreement with the fully expanded scenario 5, which treats

degradation by ozone. The Košetice site is the closest toconurbations among the EMEP sites available during 2006.This could indicate that the heterogeneous chemistry, whichis implemented as an upper estimate in this model, tendsto underestimate BaP levels further away from the sources.The best model performance was observed at the Pallas sta-tion (FI0036R), where scenario 4 was capable of matching50 % of the measurements within a factor of 2. For the re-maining remote sites in Finland, the simulation under sce-nario 5 calculated underestimated BaP levels. Based on theindex of agreement (IOA) metric, a ranking of the overallmodel performance similar to the Pallas site was observedfor the rest of the monitoring sites. Having compared site-specific performance to previous modelling efforts over Eu-rope (Aulinger et al., 2007; Matthias et al., 2009), we reachthe same conclusion as pointed out with the spatial distri-butions; i.e. we attribute the lower-performance metrics pre-sented here to the substantially weaker emission sources inrecent years, hence the better agreement with the more re-cent SMOKE-EU/CMAQ application (Bieser et al., 2012).With respect to other annual simulations, we observed loweraverage BaP levels over the Piedmont region compared tothe FARM model, which were higher than the EMEP (MSC-E) output used in the same study (Silibello et al., 2012). Fi-nally, a qualitative analysis of the deposition measurementsfor the few stations that measured BaP showed that CMAQcalculated plausible patterns of dry and wet deposition (seeFig. S10).

5 Conclusions

This study presented the development of a new modellingsystem for investigating the dynamics of transport and gas-particle partitioning schemes for POPs at the regional scaleover Europe. The implementation of this model is basedon the WRF-CMAQ model framework with additional algo-rithms for GPP schemes and a module that accounts for het-erogeneous reactivity of BaP with ozone. Predictions fromthe WRF-CMAQ-BaP model were compared to measure-ments obtained from the EMEP monitoring network. Modelevaluation has revealed satisfactory agreement to the mea-



Figure 6. Distribution of BaP mass [log cBaP (ng m−3)] in aerosols of three different modes in three different cells [urban (Vienna), suburban(CZ0003R), and remote (FI0036R)] for each scenario as calculated during a winter (20 January) and a summer (20 July) day.

surements and performance metrics similar to those of pre-vious studies with significantly higher measurement avail-ability. The results presented in this work suggest that thenew expanded model is able to simulate the ambient levels ofBaP fairly well. It is found that BaP distributions in Europeare very sensitive to the choice of GPP models. While thisstatement has been also indicated at the global scale (Lam-mel et al., 2009), GPP has not been adequately explored byprevious regional-scale studies (see Table S1 and referencestherein). We conclude that the dual OM absorption and blackcarbon adsorption (DE) parameterization (scenario 4–5) of-fer a better performance, as BaP predictions tend to be closerto the measurements. Introducing an independent upper es-timate of heterogeneous BaP reactivity is found to signifi-cantly reduce BaP levels throughout the domain, with a re-

duction pattern that follows the spatial distribution of ozoneand emission sources. Despite the limited BaP monitoringnetwork, a better agreement was observed when comple-menting WRF-CMAQ-BaP with the upper estimate for het-erogeneous reactivity (scenario 5) at certain EMEP sites. Un-certainties and limitations of the current emission inventoryestimates of BaP, OM, and BC, along with insufficient labo-ratory data, are of major concern, hampering efforts to fullyevaluate the long-range transport potential, the effect of pho-tochemistry, and interactions of such compounds. In addi-tion, certain simplifications in the treatment of coarse PM andOM within the aerosol module of CMAQ (Binkowski, 2003;Mathur et al., 2008) are posing limitations on fully exploit-ing the advanced empirical gas-particle partitioning models(Koa, Dachs-Eisenreich) in this specific mode. An implicit



Figure 7. Model predicted and observed concentration (ng m−3) time series at selected EMEP monitoring sites during 2006 for the dualGPP model scenario 4.

overlap of the empiric adsorption model (JP) with theKoa (oranother absorption) model (see Sect. 2.1.1) may artificiallyenhance partitioning, something that has not been addressedby previous modelling efforts. However, such discrepancy isexpected to be negligible for BaP, especially when comparedto existing uncertainties of output aerosol parameters prop-agating from emission estimates and from the current de-sign and state of development of the CMAQ aerosol module(AERO).

The modelling approach presented here allows a si-multaneous estimation of organic pollutants, includingsemivolatiles (such as most POPs) within the CMAQ frame-work, and can be used as a supplementary component ofa population exposure modelling system. Although similarsystems are currently explored, they rely heavily on observa-tion data while using a very basic air quality model structure(Guerreiro et al., 2016a, b). At this time, annual exposure es-timates are more useful due to the low accuracy accompany-ing highly time-resolved models and observations. As previ-

ous efforts to evaluate the use of mesoscale models for BaPdispersion studies also conclude, changes in emission pat-terns and their strengths are of particular importance. Whilethere is undeniable effort to shift from fossil fuels to renew-ables which has led to updated emission inventories, domes-tic heating across Europe has also a complex relationship toeconomic factors and human activities that varies tremen-dously between countries and fuel types (Lohmann et al.,2006; Saffari et al., 2013). Besides the need for updated emis-sion inventories for the eastern Europe, extended simulationsare suggested for years with higher measurement availabilityand density. Following modelling, studies should focus onquantifying the long-range transport potential and examiningthe hypothesis that secondary organic aerosol (SOA) may fa-cilitate PAH transport by utilizing more sophisticated aerosoland heterogeneous chemistry parameterizations or submod-els (i.e. accounting also for the burial effect; see Sect. 2.2).More specifically, recent evidence suggests that, in cold anddry air, accessibility of PAHs in OM is reduced (due to low



Figure 8. Model predicted and observed monthly BaP concentrations (ng m−3) across all EMEP sites in 2006 for the dual GPP modelscenario 4.

diffusivity), which might explain the apparent inconsistencyof high LRT potential (BaP levels in the Arctic) on onehand and the relative high heterogeneous reactivity measuredin the laboratory on the other (Friedman et al., 2014; Ze-lenyuk et al., 2012; Zhou et al., 2012, 2013). To understandthe atmospheric lifetime of PAH, interaction between PAHand SOA, and progression of the chemical composition ofambient OM, a new need for additional studies quantifyingGPP and LRT potential under a wide range of atmosphericconditions is emerging. In view of such findings, the WRF-CMAQ-BaP modelling system should be extended to studya wide range of additional organic pollutants and processes(e.g. multicompartmental cycling, biodegradation, heteroge-neous chemistry).

6 Data availability

The WRF-CMAQ-BaP output data used in this study arearchived in MetaCentrum and are available upon requestfrom the authors ([email protected]).

The Supplement related to this article is available onlineat doi:10.5194/acp-16-15327-2016-supplement.

Acknowledgements. The authors thank Frank Binkowski,Sarav Arunachalam, and Zac Adelman at the University of NorthCarolina (UNC) for their guidance and support during the onsetof this project. EMEP is acknowledged for the monitoring data.Access to computing and storage facilities owned by partiesand projects contributing to the National Grid InfrastructureMetaCentrum provided under the programme “Projects of LargeInfrastructure for Research, Development, and Innovations”(LM2010005) is also greatly appreciated. This research wassupported by the Granting Agency of the Czech Republic (projectno. 312334), the National Sustainability Programme of the CzechMinistry of Education, Youth and Sports (LO1214), and theRECETOX research infrastructure (LM2015051).

Edited by: I. RiipinenReviewed by: three anonymous referees

References

Aas, W. and Breivik, K.: Heavy metals and POP measurements,2010, EMEP/CCC, Kjeller, 2012.

Andersson, J. T. and Achten, C.: Time to say goodbye to the16 EPA PAHs? Toward an up-to-date use of PACs for envi-ronmental purposes, Polycycl. Aromat. Comp., 35, 330–354,doi:10.1080/10406638.2014.991042, 2015.


http://dx.doi.org/10.5194/acp-16-15327-2016-supplement

http://dx.doi.org/10.1080/10406638.2014.991042


Atkinson, R. and Arey, J.: Atmospheric chemistry of gas-phasepolycyclic aromatic hydrocarbons: formation of atmosphericmutagens, Environ. Health Persp., 102, 117–126, 1994.

Aulinger, A., Matthias, V., and Quante, M.: Introducing a partition-ing mechanism for PAHs into the Community Multiscale AirQuality Modeling System and its application to simulating thetransport of benzo(a)pyrene over Europe, J. Appl. Meteor. Cli-matol., 46, 1718–1730, doi:10.1175/2007JAMC1395.1, 2007.

Aulinger, A., Matthias, V., and Quante, M.: An approach to tem-porally disaggregate benzo(a)pyrene emissions and their appli-cation to a 3D Eulerian atmospheric chemistry transport model,Water Air Soil Pollut., 216, 1–13, doi:10.1007/s11270-010-0559-x, 2011.

Bärring, H., Bucheli, T. D., Broman, D., and Gustafsson,Ö.: Soot–water distribution coefficients for polychlorinateddibenzo-p-dioxins, polychlorinated dibenzofurans and poly-brominated diphenylethers determined with the soot cosolvency-column method, Chemosphere, 49, 515–523, doi:10.1016/s0045-6535(02)00389-2, 2002.

Bewersdorff, I., Aulinger, A., Matthias, V., and Quante, M.: Theeffect of temporal resolution of PAH emission data on transportand deposition patterns simulated with the Community Multi-scale Air Quality modelling system (CMAQ), Meteorol. Z., 18,41–53, doi:10.1127/0941-2948/2009/351, 2009.

Beyer, A., Mackay, D., Matthies, M., Wania, F., and Web-ster, E.: Assessing long-range transport potential of persis-tent organic pollutants, Environ. Sci. Technol., 34, 699–703,doi:10.1021/es990207w, 2000.

Bieser, J., Aulinger, A., Matthias, V., Quante, M., and Builtjes, P.:SMOKE for Europe – adaptation, modification and evaluationof a comprehensive emission model for Europe, Geosci. Model.Dev., 4, 47–68, doi:10.5194/gmd-4-47-2011, 2011.

Bieser, J., Aulinger, A., Matthias, V., and Quante, M.: Impact ofemission reductions between 1980 and 2020 on atmosphericbenzo[a]pyrene concentrations over Europe, Water Air Soil Pol-lut., 223, 1393–1414, doi:10.1007/s11270-011-0953-z, 2012.

Binkowski, F. S.: Models-3 Community Multiscale Air Quality(CMAQ) model aerosol component, 1. Model description, J.Geophys. Res., 108, 4183, doi:10.1029/2001JD001409, 2003.

Bond, T. C. and Bergstrom, R. W.: Light absorption by carbona-ceous particles: An investigative review, Aerosol. Sci. Tech., 40,27–67, doi:10.1080/02786820500421521, 2006.

Byun, D. and Schere, K. L.: Review of the governing equations,computational algorithms, and other components of the Models-3 Community Multiscale Air Quality (CMAQ) modeling system,Appl. Mech. Rev., 59, 51–77, 2006.

Byun, D. W.: Dynamically consistent formulations in meteorolog-ical and air quality models for multiscale atmospheric studies,Part II: Mass conservation issues, J. Atmos. Sci., 56, 3789–3807,doi:10.1175/1520-0469(1999)056<3789:DCFIMA>2.0.CO;2,1999a.

Byun, D. W.: Dynamically consistent formulations in meteorolog-ical and air quality models for multiscale atmospheric studies,Part II: Mass conservation issues, J. Atmos. Sci., 56, 3808–3820,doi:10.1175/1520-0469(1999)056<3808:DCFIMA>2.0.CO;2,1999b.

Byun, D. W. and Ching, J.: Science algorithms of the EPA Models-3 community multiscale air quality (CMAQ) modeling system,US Environmental Protection Agency, Office of Research and

Development Washington, DC, USA, Washington, DC 20460.,1999.

Carslaw, D. C. and Ropkins, K.: openair – An R package for airquality data analysis, Environ. Model. Softw., 27/28, 52–61,doi:10.1016/j.envsoft.2011.09.008, 2012.

Chang, J., Middleton, P., Stockwell, W., Pleim, J. E., Lansford, H.,Binkowski, F. S., Seaman, N., Staufer, D., and Madronic, S.: Theregional acid deposition model and engineering model, in: AcidDeposition State of Science and Technology, 1986.

Cooter, E. J. and Hutzell, W. T.: A regional atmosphericfate and transport model for atrazine, 1. Developmentand implementation, Environ. Sci. Technol., 36, 4091–4098,doi:10.1021/es011371y, 2002.

Dachs, J. and Eisenreich, S. J.: Adsorption onto aerosol sootcarbon dominates gas-particle partitioning of polycyclic aro-matic hydrocarbons, Environ. Sci. Technol., 34, 3690–3697,doi:10.1021/es991201, 2000.

Dachs, J., Ribes, S., van Drooge, B., Grimalt, J., Eisenreich, S.J., and Gustafsson, Ö.: Response to the comment on “Influ-ence of soot carbon on the soil-air partitioning of polycyclicaromatic hydrocarbons”, Environ. Sci. Technol., 38, 1624–1625,doi:10.1021/es0307118, 2004.

de Maagd, P., Hulscher, Ten, D., Van den Heuvel, H., Opperhuizen,A., and Sijm, D.: Physicochemical properties of polycyclic aro-matic hydrocarbons: Aqueous solubilities, n-octanol/water par-tition coefficients, and Henry’s law constants, Environ. Toxicol.Chem., 17, 251–257, 1998.

Denier van der Gon, H. A. C., Visschedijk, A. J. H., and Bolscher,M. V. H.: Study to the effectiveness of the UNECE Persistent Or-ganic Pollutants (POP), Protocol and cost of additional measures,Phase II: Estimated emission reduction and cost of options for apossible revision of the POP Protocol, TNO, 2006.

de Smet, P., Horálek, J., Conková, M., Kurfürst, P., de Leeuw, F.,and Denby, B.: European air quality maps of ozone and PM10for 2006 and their uncertainty analysis, ETAC/ACC TechnicalPaper 2008/8, 2009.

DETR: Polycyclic Aromatic Hydrocarbons, Department of the En-vironment, Transport and the Regions Expert Panel on Air Qual-ity Standards, Stationery Office, London, 2009.

Diamond, M. L. and Hodge, E.: Urban Contaminant Dynamics:From Source to Effect, Environ. Sci. Technol., 41, 3796–3800,doi:10.1021/es072542n, 2007.

EC: Directive 2004/107/EC of the European Parliament and of theCouncil, Official Journal of the European Union, 2004.

EEA: Status of black carbon monitoring in ambient air in Europe,European Environment Agency, European Environment Agency,2013.

Fernandez, P., Grimalt, J. O., and Vilanova, R. M.: Atmosphericgas-particle partitioning of polycyclic aromatic hydrocarbons inhigh mountain regions of Europe, Environ. Sci. Technol., 36,1162–1168, doi:10.1021/es010190t, 2002.

Finizio, A., Mackay, D., Bidleman, T., and Harner, T.: Octanol-airpartition coefficient as a predictor of partitioning of semi-volatileorganic chemicals to aerosols, Atmos. Environ., 31, 2289–2296,doi:10.1016/S1352-2310(97)00013-7, 1997.

Finlayson-Pitts, B. J. and Pitts Jr., J. N.: Chemistry of the Upper andLower Atmosphere, Academic Press, 1999.

Friedman, C. L. and Selin, N. E.: Long-range atmospheric transportof polycyclic aromatic hydrocarbons: a global 3-D model analy-


http://dx.doi.org/10.1175/2007JAMC1395.1

http://dx.doi.org/10.1007/s11270-010-0559-x

http://dx.doi.org/10.1007/s11270-010-0559-x

http://dx.doi.org/10.1016/s0045-6535(02)00389-2

http://dx.doi.org/10.1016/s0045-6535(02)00389-2

http://dx.doi.org/10.1127/0941-2948/2009/351

http://dx.doi.org/10.1021/es990207w

http://dx.doi.org/10.5194/gmd-4-47-2011

http://dx.doi.org/10.1007/s11270-011-0953-z

http://dx.doi.org/10.1029/2001JD001409

http://dx.doi.org/10.1080/02786820500421521

http://dx.doi.org/10.1175/1520-0469(1999)056<3789:DCFIMA>2.0.CO;2

http://dx.doi.org/10.1175/1520-0469(1999)056<3808:DCFIMA>2.0.CO;2

http://dx.doi.org/10.1016/j.envsoft.2011.09.008

http://dx.doi.org/10.1021/es011371y

http://dx.doi.org/10.1021/es991201


http://dx.doi.org/10.1021/es072542n

http://dx.doi.org/10.1021/es010190t

http://dx.doi.org/10.1016/S1352-2310(97)00013-7


sis including evaluation of arctic sources, Environ. Sci. Technol.,46, 9501–9510, doi:10.1021/es301904d, 2012.

Friedman, C. L., Pierce, J. R., and Selin, N. E.: Assessing theinfluence of secondary organic versus primary carbonaceousaerosols on long-range atmospheric polycyclic aromatic hy-drocarbon transport, Environ. Sci. Technol., 48, 3293–3302,doi:10.1021/es405219r, 2014.

Galarneau, E., Makar, P. A., Zheng, Q., Narayan, J., Zhang, J.,Moran, M. D., Bari, M. A., Pathela, S., Chen, A., and Chlum-sky, R.: PAH concentrations simulated with the AURAMS-PAHchemical transport model over Canada and the USA, Atmos.Chem. Phys., 14, 4065–4077, doi:10.5194/acp-14-4065-2014,2014.

Goss, K.-U.: Comment on “Influence of soot carbon on the soil-airpartitioning of polycyclic aromatic hydrocarbon”, Environ. Sci.Technol., 38, 1622–1623, doi:10.1021/es0301370, 2004.

Götz, C. W., Scheringer, M., MacLeod, M., Roth, C. M., andHungerbühler, K.: Alternative approaches for modeling gas-particle partitioning of semivolatile organic chemicals: modeldevelopment and comparison, Environ. Sci. Technol., 41, 1272–1278, doi:10.1021/es060583y, 2007.

Grell, G. A.: Prognostic Evaluation of Assumptions Used byCumulus Parameterizations, Mon. Weather Rev., 121, 764–787,doi:10.1175/1520-0493(1993)121<0764:PEOAUB>2.0.CO;2,1993.

Grell, G. A. and Devenyi, D.: A generalized approach toparameterizing convection combining ensemble and dataassimilation techniques, Geophys. Res. Lett., 29, 1693,doi:10.1029/2002GL015311, 2002.

Guerreiro, C., Horálek, J., and De Leeuw, F.: Benzo(a)pyrenein Europe: Ambient air concentrations, population expo-sure and health effects, Environ. Pollut., 214, 657–667,doi:10.1016/j.envpol.2016.04.081, 2016a.

Guerreiro, C., Horálek, J., de Leeuw, F., and Couvidat, F.: Mappingambient concentrations of benzo(a)pyrene in Europe, EuropeanTopic Centre on Air Pollution and Climate Change Mitigation,Technical Paper 2014/6, 2016b.

Gusev, A., Mantseva, E., and Shatalov, V.: Regional multicompart-ment model MSCE-POP, MSC-E Technical Report, 2005.

Halsall, C. J., Sweetman, A. J., Barrie, L. A., and Jones, K. C.:Modelling the behaviour of PAHs during atmospheric trans-port from the UK to the Arctic, Atmos. Environ., 35, 255–267,doi:10.1016/S1352-2310(00)00195-3, 2001.

Hansen, K. M., Halsall, C. J., and Christensen, J. H.: A dynamicmodel to study the exchange of gas-phase persistent organic pol-lutants between air and a seasonal snowpack, Environ. Sci. Tech-nol., 40, 2644–2652, doi:10.1021/es051685b, 2006.

Harner, T. and Bidleman, T. F.: Octanol-air partition coeffi-cient for describing particle/gas partitioning of aromatic com-pounds in urban air, Environ. Sci. Technol., 32, 1494–1502,doi:10.1021/es970890r, 1998.

Iacono, M. J., Delamere, J. S., Mlawer, E. J., Shephard, M.W., Clough, S. A., and Collins, W. D.: Radiative forcing bylong-lived greenhouse gases: Calculations with the AER radia-tive transfer models, J. Geophys. Res-Atmos., 113, D13103,doi:10.1029/2008JD009944, 2008.

Inomata, Y., Kajino, M., Sato, K., Ohara, T., Kurokawa, J.-I.,Ueda, H., Tang, N., Hayakawa, K., Ohizumi, T., and Aki-moto, H.: Emission and atmospheric transport of particulate

PAHs in Northeast Asia, Environ. Sci. Technol., 46, 4941–4949,doi:10.1021/es300391w, 2012.

Jonker, M. T. O. and Koelmans, A. A.: Sorption of polycyclicaromatic hydrocarbons and polychlorinated biphenyls to sootand soot-like materials in the aqueous environment: Mecha-nistic considerations, Environ. Sci. Technol., 36, 3725–3734,doi:10.1021/es020019x, 2002.

Junge, C. E.: Fate of pollutants in the air and water environment,edited by: Suffet, I. H., Wiley, New York, 1977.

Kahan, T. F. and Donaldson, D. J.: Heterogeneous ozonation kinet-ics of polycyclic aromatic hydrocarbons on organic films, Atmos.Environ., 40, 3448–3459, doi:10.1016/j.atmosenv.2006.02.004,2006.

Kaiser, J. C., Riemer, N., and Knopf, D. A.: Detailed heterogeneousoxidation of soot surfaces in a particle-resolved aerosol model,Atmos. Chem. Phys., 11, 4505–4520, doi:10.5194/acp-11-4505-2011, 2011.

Kalberer, M., Paulsen, D., Sax, M., Steinbacher, M., Dommen, J.,Prevot, A. S. H., Fisseha, R., Weingartner, E., Frankevich, V.,Zenobi, R., and Baltensperger, U.: Identification of polymersas major components of atmospheric organic aerosols, Science,303, 1659–1662, doi:10.1126/science.1092185, 2004.

Keyte, I. J., Harrison, R. M., and Lammel, G.: Chemical reac-tivity and long-range transport potential of polycyclic aromatichydrocarbons – a review, Chem. Soc. Rev., 42, 9333–9391,doi:10.1039/C3CS60147A, 2013.

Kim, Y. P., Seinfeld, J. H., and Saxena, P.: Atmospheric gas-aerosolequilibrium, I. Thermodynamic model, Aerosol Sci. Tech., 19,157–181, doi:10.1080/02786829308959628, 1993.

Kwamena, N.-O. A., Thornton, J. A., and Abbatt, J. P. D.: Ki-netics of surface-bound benzo[a]pyrene and ozone on solid or-ganic and salt aerosols, J. Phys. Chem. A, 108, 11626–11634,doi:10.1021/jp046161x, 2004.

Kwamena, N.-O. A., Clarke, J. P., Kahan, T. F., Diamond, M. L., andDonaldson, D. J.: Assessing the importance of heterogeneous re-actions of polycyclic aromatic hydrocarbons in the urban atmo-sphere using the Multimedia Urban Model, Atmos. Environ., 41,37–50, doi:10.1016/j.atmosenv.2006.08.016, 2007.

Lammel, G., Sehili, A. M., Bond, T. C., Feichter, J., and Grassl,H.: Gas/particle partitioning and global distribution of polycyclicaromatic hydrocarbons – A modelling approach, Chemosphere,76, 98–106, doi:10.1016/j.chemosphere.2009.02.017, 2009.

Lammel, G., Klánová, J., Ilic, P., Kohoutek, J., Gasic, B., Kovacic,I., and Škrdlíková, L.: Polycyclic aromatic hydrocarbons in airon small spatial and temporal scales – II. Mass size distributionsand gas-particle partitioning, Atmos. Environ., 44, 5022–5027,doi:10.1016/j.atmosenv.2010.08.001, 2010.

Lammel, G., Heil, A., Stemmler, I., Dvorská, A., and Klánová,J.: On the contribution of biomass burning to POPs (PAHs andPCDDs) in air in Africa, Environ. Sci. Technol., 47, 11616–11624, doi:10.1021/es401499q, 2013.

Lang, C., Tao, S., Zhang, G., Fu, J., and Simonich, S.:Outflow of polycyclic aromatic hydrocarbons from Guang-dong, Southern China, Environ. Sci. Technol., 41, 8370–8375,doi:10.1021/es071853v, 2007.

Lang, C., Tao, S., Liu, W., Zhang, Y., and Simonich, S.: At-mospheric transport and outflow of polycyclic aromatic hydro-carbons from China, Environ. Sci. Technol., 42, 5196–5201,doi:10.1021/es800453n, 2008.


http://dx.doi.org/10.1021/es301904d

http://dx.doi.org/10.1021/es405219r

http://dx.doi.org/10.5194/acp-14-4065-2014


http://dx.doi.org/10.1021/es060583y

http://dx.doi.org/10.1175/1520-0493(1993)121<0764:PEOAUB>2.0.CO;2

http://dx.doi.org/10.1029/2002GL015311

http://dx.doi.org/10.1016/j.envpol.2016.04.081

http://dx.doi.org/10.1016/S1352-2310(00)00195-3

http://dx.doi.org/10.1021/es051685b

http://dx.doi.org/10.1021/es970890r

http://dx.doi.org/10.1029/2008JD009944

http://dx.doi.org/10.1021/es300391w

http://dx.doi.org/10.1021/es020019x

http://dx.doi.org/10.1016/j.atmosenv.2006.02.004



http://dx.doi.org/10.1126/science.1092185

http://dx.doi.org/10.1039/C3CS60147A

http://dx.doi.org/10.1080/02786829308959628

http://dx.doi.org/10.1021/jp046161x


http://dx.doi.org/10.1016/j.chemosphere.2009.02.017


http://dx.doi.org/10.1021/es401499q

http://dx.doi.org/10.1021/es071853v

http://dx.doi.org/10.1021/es800453n


Liu, S., Tao, S., Liu, W., Liu, Y., Dou, H., Zhao, J., Wang, L., Wang,J., Tian, Z., and Gao, Y.: Atmospheric polycyclic aromatic hy-drocarbons in North China: A winter-time study, Environ. Sci.Technol., 41, 8256–8261, doi:10.1021/es0716249, 2007.

Lohmann, R. and Lammel, G.: Adsorptive and absorptive con-tributions to the gas-particle partitioning of polycyclic aro-matic hydrocarbons: state of knowledge and recommendedparametrization for modeling, Environ. Sci. Technol., 38, 3793–3803, doi:10.1021/es035337q, 2004.

Lohmann, R., Lee, R. G. M., Abbott, J., Coleman, P., and Jones, K.C.: Verifying emission factors and national POPs emission inven-tories for the UK using measurements and modelling at two rurallocations, J. Environ. Monit., 8, 79–88, doi:10.1039/B511399G,2006.

Mantseva, E., Dutchak, S., Rozovskaya, O., and Shatalov, V.: EMEPcontribution to the preparatory work for the review of the CLR-TAP Protocol on Persistent Organic Pollutants, EMEP MSC-EInformation Note, 2004.

Mathur, R., Yu, S., Kang, D., and Schere, K. L.: Assess-ment of the wintertime performance of developmental partic-ulate matter forecasts with the Eta-Community Multiscale AirQuality modeling system, J. Geophys. Res., 113, D0230315,doi:10.1029/2007JD008580, 2008.

Matthias, V., Aulinger, A., and Quante, M.: Adapting CMAQ toinvestigate air pollution in north sea coastal regions, Environ.Model. Softw., 23, 356–368, doi:10.1016/j.envsoft.2007.04.010,2008.

Matthias, V., Aulinger, A., and Quante, M.: CMAQ sim-ulations of the benzo(a)pyrene distribution over Europefor 2000 and 2001, Atmos. Environ., 43, 4078–4086,doi:10.1016/j.atmosenv.2009.04.058, 2009.

Meng, F., Zhang, B., Gbor, P., Wen, D., Yang, F., Shi, C.,Aronson, J., and Sloan, J.: Models for gas/particle partition-ing, transformation and air/water surface exchange of PCBsand PCDD/Fs in CMAQ, Atmos. Environ., 41, 9111–9127,doi:10.1016/j.atmosenv.2007.08.009, 2007.

Milbrandt, J. A. and Yau, M. K.: A multimoment bulk mi-crophysics parameterization, Part I: Analysis of the role ofthe spectral shape parameter, J. Atmos. Sci., 62, 3051–3064,doi:10.1175/JAS3535.1, 2005a.

Milbrandt, J. A. and Yau, M. K.: A multimoment bulk micro-physics parameterization, Part II: A proposed three-moment clo-sure and scheme description, J. Atmos. Sci., 62, 3065–3081,doi:10.1175/JAS3535.1, 2005b.

Milford, C., Fernández-Camacho, R., la Campa, de, A. M. S., Ro-dríguez, S., Castell, N., Marrero, C., Bustos, J. J., la Rosa, de,J. D., and Stein, A. F.: Black Carbon aerosol measurements andsimulation in two cities in south-west Spain, Atmos. Environ.,126, 55–65, doi:10.1016/j.atmosenv.2015.11.026, 2016.

Nenes, A., Pandis, S. N., and Pilinis, C.: ISORROPIA: Anew thermodynamic equilibrium model for multiphase multi-component inorganic aerosols, Aquat. Geochem., 4, 123–152,doi:10.1023/A:1009604003981, 1998.

Ngabe, B. and Poissant, L.: Polycyclic aromatic hydrocarbons in theair in the St. Lawrence Basin (Quebec), Environ. Sci. Technol.,37, 2094–2099, doi:10.1021/es0210345, 2003.

Odabasi, M., Cetin, E., and Sofuoglu, A.: Determination of octanol–air partition coefficients and supercooled liquid vapor pressuresof PAHs as a function of temperature: Application to gas–particle

partitioning in an urban atmosphere, Atmos. Environ., 40, 6615–6625, doi:10.1016/j.atmosenv.2006.05.051, 2006.

Offenberg, J. H. and Baker, J. E.: Aerosol size distributionsof polycyclic aromatic hydrocarbons in urban and over-water atmospheres, Environ. Sci. Technol., 33, 3324–3331,doi:10.1021/es990089c, 1999.

Otte, T. L. and Pleim, J. E.: The Meteorology-Chemistry Inter-face Processor (MCIP) for the CMAQ modeling system: up-dates through MCIPv3.4.1, Geosci. Model Dev., 3, 243–256,doi:10.5194/gmd-3-243-2010, 2010.

Pacyna, J. M., Breivik, K., Münch, J., and Fudala, J.: Europeanatmospheric emissions of selected persistent organic pollutants,1970–1995, Atmos. Environ., 37, 119–131, doi:10.1016/S1352-2310(03)00240-1, 2003.

Pankow, J. F.: Review and comparative-analysis of the theo-ries on partitioning between the gas and aerosol particulatephases in the atmosphere, Atmos. Environ., 21, 2275–2283,doi:10.1016/0004-6981(87)90363-5, 1987.

Pankow, J. F.: An absorption model of gas/particle partitioningof organic compounds in the atmosphere, Atmos. Environ., 28,185–188, doi:10.1016/1352-2310(94)90093-0, 1994.

Pankow, J. F.: Further discussion of the octanol/air partitioncoefficient Koa as a correlating parameter for gas/particlepartitioning coefficients, Atmos. Environ., 32, 1493–1497,doi:10.1016/S1352-2310(97)00383-X, 1998.

Pankow, J. F.: An absorption model of the gas/aerosol partitioninginvolved in the formation of secondary organic aerosol, Atmos.Environ., 41, 75–79, doi:10.1016/j.atmosenv.2007.10.060, 2007.

Pedersen, D. U., Durant, J. L., Penman, B. W., Crespi, C. L.,Hemond, H. F., Lafleur, A. L., and Cass, G. R.: Human cellmutagens in respirable airborne particles from the NortheasternUnited States, 1. Mutagenicity of fractionated samples, Environ.Sci. Technol., 38, 682–689, doi:10.1021/es0347282, 2004.

Pedersen, D. U., Durant, J. L., Taghizadeh, K., Hemond, H. F.,Lafleur, A. L., and Cass, G. R.: Human cell mutagens in res-pirable airborne particles from the Northeastern United States,2. Quantification of mutagens and other organic compounds,Environ. Sci. Technol., 39, 9547–9560, doi:10.1021/es050886c,2005.

Perraudin, E., Budzinski, H., and Villenave, E.: Kinetic study ofthe reactions of ozone with polycyclic aromatic hydrocarbonsadsorbed on atmospheric model particles, J. Atmos. Chem., 56,57–82, doi:10.1007/s10874-006-9042-x, 2007.

Pleim, J. E.: A combined local and nonlocal closure modelfor the atmospheric boundary layer, Part I: Model descrip-tion and testing, J. Appl. Meteor. Climatol., 46, 1383–1395,doi:10.1175/JAM2539.1, 2007.

Pleim, J. E. and Xiu, A.: Development and testing of a surface fluxand planetary boundary layer model for application in mesoscalemodels, J. Appl. Meteor., 34, 16–32, doi:10.1175/1520-0450(1995)034<0034:DATOAS>2.0.CO;2, 1995.

Pleim, J. E. and Xiu, A. J.: Development of a land surface model,Part II: Data assimilation, J. Appl. Meteor., 42, 1811–1822,doi:10.1175/1520-0450(2003)042<1811:DOALSM>2.0.CO;2,2003.

Pleim, J. E., Finkelstein, P. L., Clarke, J. F., and Ellestad, T. G.:A technique for estimating dry deposition velocities based onsimilarity with latent heat flux, Atmos. Environ., 33, 2257–2268,doi:10.1016/S1352-2310(98)00162-9, 1999.



http://dx.doi.org/10.1021/es035337q

http://dx.doi.org/10.1039/B511399G

http://dx.doi.org/10.1029/2007JD008580

http://dx.doi.org/10.1016/j.envsoft.2007.04.010



http://dx.doi.org/10.1175/JAS3535.1

http://dx.doi.org/10.1175/JAS3535.1


http://dx.doi.org/10.1023/A:1009604003981



http://dx.doi.org/10.1021/es990089c

http://dx.doi.org/10.5194/gmd-3-243-2010

http://dx.doi.org/10.1016/S1352-2310(03)00240-1

http://dx.doi.org/10.1016/S1352-2310(03)00240-1

http://dx.doi.org/10.1016/0004-6981(87)90363-5

http://dx.doi.org/10.1016/1352-2310(94)90093-0

http://dx.doi.org/10.1016/S1352-2310(97)00383-X



http://dx.doi.org/10.1021/es050886c

http://dx.doi.org/10.1007/s10874-006-9042-x

http://dx.doi.org/10.1175/JAM2539.1

http://dx.doi.org/10.1175/1520-0450(1995)034<0034:DATOAS>2.0.CO;2

http://dx.doi.org/10.1175/1520-0450(1995)034<0034:DATOAS>2.0.CO;2

http://dx.doi.org/10.1175/1520-0450(2003)042<1811:DOALSM>2.0.CO;2

http://dx.doi.org/10.1016/S1352-2310(98)00162-9


Pöschl, U., Letzel, T., Schauer, C., and Niessner, R.: Interactionof ozone and water vapor with spark discharge soot aerosolparticles coated with benzo[a]pyrene: O3 and H2O adsorption,benzo[a]pyrene degradation, and atmospheric implications, J.Phys. Chem. A, 105, 4029–4041, doi:10.1021/jp004137n, 2001.

Prevedouros, K., Jones, K. C., and Sweetman, A. J.: Mod-elling the atmospheric fate and seasonality of polycyclic aro-matic hydrocarbons in the UK, Chemosphere, 56, 195–208,doi:10.1016/j.chemosphere.2004.02.032, 2004.

Prevedouros, K., Palm-Cousins, A., Gustafsson, Ö., and Cousins, I.T.: Development of a black carbon-inclusive multi-media model:Application for PAHs in Stockholm, Chemosphere, 70, 607–615,doi:10.1016/j.chemosphere.2007.07.002, 2008.

Saffari, A., Daher, N., Samara, C., Voutsa, D., Kouras, A., Manoli,E., Karagkiozidou, O., Vlachokostas, C., Moussiopoulos, N.,Shafer, M. M., Schauer, J. J., and Sioutas, C.: Increased biomassburning due to the economic crisis in Greece and its adverseimpact on wintertime air quality in Thessaloniki, Environ. Sci.Technol., 47, 13313–13320, doi:10.1021/es403847h, 2013.

San José, R., Pérez, J. L., Callén, M. S., López, J. M., and Mas-tral, A.: BaP (PAH) air quality modelling exercise over Zaragoza(Spain) using an adapted version of WRF-CMAQ model, En-viron. Pollut., 183, 151–158, doi:10.1016/j.envpol.2013.02.025,2013.

Sehili, A. M. and Lammel, G.: Global fate and distribu-tion of polycyclic aromatic hydrocarbons emitted fromEurope and Russia, Atmos. Environ., 41, 8301–8315,doi:10.1016/j.atmosenv.2007.06.050, 2007.

Shatalov, V., Gusev, A., Dutchak, S., Holoubek, I., Mantseva, E.,Rozovskaya, O., Sweetman, A., Strukov, B., and Vulykh, N.:Modelling of POP contamination in European region: evaluationof the model performance, Moscow, 2005.

Shiraiwa, M., Garland, R. M., and Pöschl, U.: Kinetic double-layermodel of aerosol surface chemistry and gas-particle interactions(K2-SURF): Degradation of polycyclic aromatic hydrocarbonsexposed to O3, NO2, H2O, OH and NO3, Atmos. Chem. Phys.,9, 9571–9586, doi:10.5194/acp-9-9571-2009, 2009.

Shiraiwa, M., Pfrang, C., and Pöschl, U.: Kinetic multi-layer modelof aerosol surface and bulk chemistry (KM-SUB): the influ-ence of interfacial transport and bulk diffusion on the oxidationof oleic acid by ozone, Atmos. Chem. Phys., 10, 3673–3691,doi:10.5194/acp-10-3673-2010, 2010.

Silibello, C., Calori, G., Costa, M. P., Dirodi, M. G., Mircea,M., Radice, P., Vitali, L., and Zanini, G.: Benzo[a]pyrenemodelling over Italy: comparison with experimental dataand source apportionment, Atmos. Pollut. Res., 3, 399–407,doi:10.5094/APR.2012.046, 2012.

Springmann, M., Knopf, D. A., and Riemer, N.: Detailed het-erogeneous chemistry in an urban plume box model: re-versible co-adsorption of O3, NO2, and H2O on soot coatedwith benzo[a]pyrene, Atmos. Chem. Phys., 9, 7461–7479,doi:10.5194/acp-9-7461-2009, 2009.

Šrám, R., Dostal, M., Libalova, H., Rossner, P., Rossnerova, A.,Svecova, V., Topinka, J., and Bartonova, A.: The European hotspot of B[a]P and PM2.5 exposure – The Ostrava region, CzechRepublic: Health research results, ISRN Public Health, 57, 12,doi:10.3109/19396368.2010.500440, 2013.

Stokes, R. H. and Robinson, R. A.: Interactions in aqueous nonelec-trolyte solutions, I. solute-solvent equilibria, J. Phys. Chem., 70,2126–2131, doi:10.1021/j100879a010, 1966.

Van Jaarsveld, J., Van Pul, W., and De Leeuw, F.: Mod-elling transport and deposition of persistent organic pollu-tants in the European region, Atmos. Environ., 31, 1011–1024,doi:10.1016/S1352-2310(96)00251-8, 1997.

van Noort, P.: A thermodynamics-based estimation model for ad-sorption of organic compounds by carbonaceous materials in en-vironmental sorbents, Environ. Toxicol. Chem., 22, 1179–1188,doi:10.1002/etc.5620220601, 2003.

Wenborn, M. J., Coleman, P. J., Passant, N. R., Lymberidi,E., Sully, J., and Weir, A.: Speciated PAH inventory forthe United Kingdom, Report AEAT 3512, available at:https://uk-air.defra.gov.uk/assets/documents/reports/cat08/0512011419_REPFIN_all_nov.pdf (last access: 6 December2016), 1999.

WHO: World Health Organization: Health risks of persistent or-ganic pollutants from long-range transboundary air pollution,WHO Regional Office for Europe, Copenhagen, 2003.

Xiu, A. J. and Pleim, J. E.: Development of a land sur-face model, Part I: Application in a mesoscale meteorologi-cal model, J. Appl. Meteor., 40, 192–209, doi:10.1175/1520-0450(2001)040<0192:DOALSM>2.0.CO;2, 2001.

Yarwood, G., Rao, S., Yocke, M., and Whitten, G.: Updates to thecarbon bond chemical mechanism: CB05, Final report to the USEPA, RT-0400675, available at http://www.camx.com. (last ac-cess: 6 December 2016), 2005.

Zelenyuk, A., Imre, D., Beránek, J., Abramson, E., Wilson,J., and Shrivastava, M.: Synergy between secondary or-ganic aerosols and long-range transport of polycyclic aro-matic hydrocarbons, Environ. Sci. Technol., 46, 12459–12466,doi:10.1021/es302743z, 2012.

Zhang, B., Meng, F., Shi, C., Yang, F., Wen, D., Aronsson, J.,Gbor, P. K., and Sloan, J. J.: Modeling the atmospheric transportand deposition of polychlorinated dibenzo-p-dioxins and diben-zofurans in North America, Atmos. Environ., 43, 2204–2212,doi:10.1016/j.atmosenv.2009.01.004, 2009.

Zhang, Y., Seigneur, C., Seinfeld, J. H., Jacobson, M., Clegg, S.L., and Binkowski, F. S.: A comparative review of inorganicaerosol thermodynamic equilibrium modules: similarities, dif-ferences, and their likely causes, Atmos. Environ., 34, 117–137,doi:10.1016/S1352-2310(99)00236-8, 2000.

Zhou, S., Lee, A. K. Y., McWhinney, R. D., and Abbatt, J. P. D.:Burial effects of organic coatings on the heterogeneous reactivityof particle-borne benzo[a]pyrene (BaP) toward ozone, J. Phys.Chem. A, 116, 7050–7056, doi:10.1021/jp3030705, 2012.

Zhou, S., Shiraiwa, M., McWhinney, R. D., Pöschl, U., and Abbatt,J. P. D.: Kinetic limitations in gas-particle reactions arising fromslow diffusion in secondary organic aerosol, Faraday Discuss.,165, 391–406, doi:10.1039/c3fd00030c, 2013.

Zhu, Y., Tao, S., Price, O. R., Shen, H., Jones, K. C., and Sweetman,A. J.: Environmental distributions of benzo[a]pyrene in China:Current and future emission reduction scenarios explored using aspatially explicit multimedia fate model, Environ. Sci. Technol.,49, 13868–13877, doi:10.1021/acs.est.5b00474, 2015.


http://dx.doi.org/10.1021/jp004137n



http://dx.doi.org/10.1021/es403847h

http://dx.doi.org/10.1016/j.envpol.2013.02.025




http://dx.doi.org/10.5094/APR.2012.046


http://dx.doi.org/10.3109/19396368.2010.500440

http://dx.doi.org/10.1021/j100879a010

http://dx.doi.org/10.1016/S1352-2310(96)00251-8

http://dx.doi.org/10.1002/etc.5620220601

https://uk-air.defra.gov.uk/assets/documents/reports/cat08/0512011419_REPFIN_all_nov.pdf

https://uk-air.defra.gov.uk/assets/documents/reports/cat08/0512011419_REPFIN_all_nov.pdf



http://www.camx.com.

http://dx.doi.org/10.1021/es302743z


http://dx.doi.org/10.1016/S1352-2310(99)00236-8

http://dx.doi.org/10.1021/jp3030705

http://dx.doi.org/10.1039/c3fd00030c

http://dx.doi.org/10.1021/acs.est.5b00474

evaluation of gas-particle partitioning in a regional air quality model

Documents